Recent results have led many to propose a microenvironment-dependent model for initiation of migratory and disseminating tumor cell behavior at both the primary tumor and within target organs that is not stably specified by genetic mutation and that is transient in time and space. This view is called the microenvironment model of metastasis. The testing of this model has been hampered in part by the lack of high-resolution in vivo microscopy methods and genetically-encoded fluorescent probes for tumor deep-tissue imaging that allow definitive identification of the microenvironments involved in initiating the migratory and disseminating tumor cell phenotype. Equally problematic are the limitations of standard analyses of expression profiles. Standard analysis of expression profiles in cancer involves identifying consistently up- and down- regulated genes. While these techniques are likely to identify sets of genes directly within affected networks, our previous theoretical results have shown that major perturbations (of which cancer is one) cause expression changes far beyond the pathway involved. Crucially, these more distant changes will be highly variable depending on the genetic background, thus tumor expression profiles are expected to be greatly dissimilar between individuals. Using this hypothesis we propose a novel systems-level analysis of cancer (SLAC), which identifies key genes based upon increase in expression variability, and which in turn offers the possibility of discovering highly non-intuitive pathway interactions connected with microenvironment regulation of breast cancer progression. By combining the multiphoton high-resolution microscopy having the wide range of excitation wavelengths with the proposed multicolor far-red fluorescent probes as versatile as conventional GFP we will advance deep-tissue cell labeling and imaging of tumor cells dynamics in vivo. This approach will make possible the intravital imaging of simultaneously up to six genetically-encoded colors in tumor studies. This in turn will provide a way to discriminate and subsequently isolate the tumor cells of multiple metastatic phenotypes based on the fluorescent color-encoded expression patterns. By correlating the behavior and fate of migrating and disseminating tumor cells obtained by the multiphoton imaging at a single-cell level with SLAC analysis of expression profiles of these cells, we will identify the key genes driving tumor cell behaviors involved in metastasis such as cell migration and dissemination.